Failover method in a clustered computer system

Abstract

In a high available clustered computer system, a standby computer starts up takeover preparation procedures at a time when occurrence of a failure is watched in an active computer and failover procedures will be started up from the takeover preparation completion status when the failure is detected in the active computer. Thus, speeding up of failover preparation procedures is realized. With such arrangement, it is possible to realize failover procedures by the hot-standby procedures when a failure occurs in an active computer.

1. A failover method for a clustered computer system including an active computer, a plurality of computers which could be standby computers to take over a procedure of said active computer, and a disk drive which is coupled to these computers and stores a file required for an application to execute the procedure of said active computer, said method comprising the steps of:
 determining a standby computer which is supposed to take over a procedure, with the fact that occurrence of a failure is detected in said active computer as being a first trigger, and notifying the determined computer; 
 allowing the notified computer to execute the takeover preparation procedure; and 
 allowing the computer which executed said takeover preparation procedure to take over a procedure that was executed by said active computer, with the fact that a failover procedure becomes necessary due to the failure in said active computer as being a second trigger. 
 
     
     
       2. A failover method according to  claim 1 , wherein said takeover preparation procedure implies a procedure to read an application program that was running on said active computer from said disk drive. 
     
     
       3. A failover method according to  claim 1 , wherein, based on said first trigger, said disk drive executes a preparation process for failover. 
     
     
       4. A failover method according to  claim 3 , wherein the preparation process for failover executed by said disk drive includes a procedure of prefetching data to be accessed during the failover procedure to a cache from a disk. 
     
     
       5. A failover method for a clustered computer system including an active computer, a plurality of computers which could be standby computers to take over a procedure of said active computer, and a disk drive which is coupled to these computers and stores a file required for an application to execute a procedure of said active computer, said method comprising the steps of:
 determining a standby computer which is supposed to take over procedures, with the fact that occurrence of a failure is detected in said active computer as being a trigger, and allowing said standby computer to execute the takeover preparation procedures; and 
 canceling the takeover preparation procedure of said standby computer, with the fact that a failover procedure becomes unnecessary due to recovery of the failure in said active computer as being a trigger.